Hi @Eirik_Hanssen,
Sorry for the confusion and thanks for the explanation. We made a decision a couple versions back that when a user is assigned to a submission in any capacity, we will grant them permissions according to their assigned role.
This accounts for all kinds of tricky situations where a journal manager/editor may also act as author or reviewer. It does introduce some confusion, but we felt it was important to minimize the exposure of blind reviews, even if a manager acting as author/reviewer is a discouraged practice. It also provides a clearer record in the activity log. If a user assigned in one role wants to perform an action in another role, they need to assign themselves in that role, and that assignment is then tracked.
In your case, I would advise against deleting submissions. If an author makes a new submission when they should submit revisions, you should either request they use the appropriate revision submission area or manually move the revisions over to the original submission. By deleting the original submission, you will lose all activity logs, which can be important when allegations of misconduct arise and an audit is requested of the full review process.
In addition, I’ll mention that we’re planning over the next year or two to provide more features for monitoring editorial stats, such as time to review, number of review rounds, ratio of submissions that make it through review, etc. If you are regularly deleting submissions, rather than archiving them, you will skew the data.